Investigating the Power of Artificial Intelligence Algorithms in Predicting Mortality Rates in Patients with Gastrointestinal Malignancies in ICU

This study aimed to investigate the efficacy of artificial intelligence algorithms in predicting mortality among patients with gastrointestinal malignancie.. In this retrospective cohort study, all the files of patients with gastrointestinal malignancies hospitalized in the intensive care and surgery departments of Imam Khomeini Hospital in Ahvaz from ۲۰۱۳-۲۰۱۸ were reviewed, and ۲۰۰ patients met the inclusion criteria. Data on laboratory test results, clinical information, and hospitalization outcomes in the intensive care unit was collected. The model was presented utilizing artificial intelligence, a genetic algorithm, and the nearest neighbor method. The artificial intelligence algorithm demonstrated a diagnostic accuracy of ۹۰%, a specificity of ۹۱.۶۷%, and a sensitivity of ۸۳.۳۴% for mortality. The genetic algorithm assigned a high weight to the following variables: gastrointestinal cancer type and hematocrit (۰.۹۸), illness status upon admission to the intensive care unit (۰.۹۴), bicarbonate rate (۰.۸۷), background infection and body temperature (۰.۸۶), duration of hospitalization (۰.۸۴), and CRP (۰.۸۳). These variables were significant and influential in determining mortality. Genetic algorithms are highly effective in predicting the mortality of patients with gastrointestinal malignancies hospitalized in intensive care units.

کلیدواژه ها:

Gastrointestinal Malignancies ، Artificial Intelligence Algorithms ، Mortality Prediction ، Intensive Care Unit